Virtue AI Documentation
Virtue AI - The AI-Native Security & Compliance Platform for AI Agents, Models, and Apps. Explore guides, references, and tutorials for our red-teaming, guardrail, and agent products.
Product Highlights
A quick look at what makes each Virtue product distinctive.
100+
Proprietary red-teaming algorithms and skills.
AgentSuite-Red50+ envs in 14+ domains
First agent-simulated environments for high-stakes red-teaming.
AgentSuite-RedRed-teaming agent
First fully automatic agent red-teaming with varied attack injection points.
VirtueGuardMultimodal · Multilingual
First real-time guardrail — covering text, code, image, video, and audio across 100+ languages.
VirtueGuardReal-time · Low false positives
Strong policy-following served by Virtue's in-house Prelude inference framework.
AgentSuite-BlueDefense in-depth
Comprehensive protection — static scanning plus low-latency real-time Guardrail.
AgentSuite-BlueCompliance and governance
Real-time guardrails with policy enforcement, full observability, and Shadow AI detection.
Virtue PlatformFaster, Token-efficient
Purpose-built inference framework — faster than vLLM and SGLang, with broad open-source model support.
Virtue PlatformIntegrations
Easy integration with Web UI Apps, ADKs (e.g., OpenAI, Google, and LangChain), local agents (e.g., OpenClaw, Claude Code).
Virtue PlatformOn-prem
Fully on-prem deployment enabled by Virtue's in-house technical stack (models, algorithms, and infra).
Product Roadmap
Pick a product to dive into its functionality, setup, integration, and deployment.
VirtueRed
Continuous red-teaming for AI applications. Uncover hidden risks, validate behavior over time, and generate audit-ready evidence for security and compliance reviews.
Read documentationAgentSuite-Red
Evaluation and red-teaming for AI agents. 14 high-stakes domains and 40+ sandboxed environments with automated judges and an adversarial agent that probes deployed agents under direct and indirect prompt-injection threat models.
Read documentationVirtueGuard
Real-time safety models and policy guardrails across text, image, video, audio, code, and custom policies.
Read documentationAgentSuite-Blue
End-to-end protection for agentic systems built on defense-in-depth principles: static MCP and Skill Guard, runtime Prompt Guard and Action Guard, fine-grained access control, full observability, and Shadow AI detection.
Read documentationFAQ
Common questions across products, deployment, and security posture. Read the FAQ